Compression And Expansion

Hello and welcome to the Sound On Sound recording and mixing podcast channel. I am Eddie Bazil. In this podcast, I'm going to delve into the world of compression and expansion and explore some creative ways to process sounds using these two powerful processors. But before I fire up the DAW and get funky, let's define what each processor is. Compressors, be they downward or upward, reduce or narrow the overall dynamic range of the audio signal. Whereas expanders, downward and upward, yes, you guessed it, do the exact opposite and extend the dynamic range. How each achieves this depends very much on the processor's topology or design and the type or mode used. A downward compressor reduces the level above the threshold, whereas an upward compressor raises the level below the threshold. Therefore, both narrow or reduce the dynamic range of the audio signal. Downward expanders reduce the level of an audio signal below the threshold, making quiet sounds quieter. This extends the dynamic range of the audio signal. Upward compressors boost the level of an audio signal above the threshold, making loud sounds even louder. This also extends the dynamic range of the audio signal. Let's start with compression and in particular compressor behavioral types or forms. Feed forward and feedback. In a feed forward compressor, the control circuit receives the audio signal before the signal goes through the amplification system. In a feedback compressor, the control signal receives the audio signal After it has passed through the amplification stage. If you require a more in depth explanation of these two forms, then head over to Sound On Sound magazine and dig up the wonderful article by Paul White on this very subject entitled Compressor Topology, Feedforward or Feedback. Ultimately, which form you choose comes down to aims and associated functionality. And if I were arm wrestled into subjectively categorizing both forms, then I'd say Feedback compression offers a smoother and more coloured response, whereas Feedforward grabs transients faster and can therefore highlight peaks in a more controlled and punchy manner. But my advice is to try an assortment of different compressor topologies. Engage if the results suit your aims. Now that we have the behavioural aspects of compression out of the way, let's look at the various common topologies on offer. This is actually more important than compressor behaviour as choosing the right type of compressor for the task at hand is critical in how the audio processing goal is realised. VCA Uses a voltage control amplifier. VCAs are fast and have low distortion, which makes them come across as clean and transparent. However, VCAs can sound aggressive if pushed, because VCAs don't add colour to a sound. They can be used for transparent compression as they sound clean and smooth. Most common VCAs used are the SSL GBU and the DBX one 60 F, also known as FET uses field effect transistors. FET compressors emulate the tube valve sound using transistor circuits. They're fast and bright sounding. FET compressors add both punch and colour to a sound and can be extremely useful for creating aggressive transient rich audio. The most famous FET compressor is the URI 1176. Opto uses a photocell as a detector and a light bulb to determine the gain reduction. Different light sensor types and illumination sources affect the gain reduction in different ways. The time lag between the photocell and the detector makes for a slow attack and release. And this makes this type of compressor perfect for processing vocals, gluing sounds and performing gentle automatic gain changes. The most famous optocompressors are the LA-2A2 base leveling amp and the mighty class A Avalon AD2044 Vari MU. Vari, which stands for variable and MU, which is an engineering term for gain, employs a valve or tube as the variable gain reduction. Although they boast very fast attack times and can therefore act as limiters, they are slow to react, thanks to all the valves and transformers, which makes them perfect for gluing purposes. And the ratio increases with gain reduction, which gives this topology such a distinct and unique sound. This type of compressor works best on the master and group buses as it glues sounds together into a single homogenous entity. Due to its smooth and transparent response, they are favourites with mastering engineers for adding that final fairy dust. They can also be used to widen and enlarge sounds within the stereo sound stage thanks to unlinked left and right processing and mid side capabilities. The most famous Vari MU compressor is the Fairchild 670. The god of all compressors. With software, not only can all these various topologies be faithfully emulated, but some plugins allow the user access to multiple topologies within a single processor. And it doesn't end there. Equalization, effects, and so on, are also supplied as additional processes to help craft textures within a single plug-in. All we need now is the plug-in to make coffee and warm the bath and we're all set to go. Okay, so let's start with some sonic mangling exercises. Let's start with a couple of compression examples. In example one, I'm using Boss Digital's Manic Compressor to process a drum beat into an exciting new texture. I'm using one drum beat to shape another drum beat. I'll explain how the routing setup works in a minute. First, the dry versions. And now, the processed versions. Boss Digital's Manic Compressor is a versatile compressor offering multiple colour or model modes to help shape the signal with emphasis on working in parallel. The models range from clean and transparent to gritty and quite frankly offensive and it is with these detailed colours in mind that the compressor was designed to shine in parallel mode. This beast comes laden with a detail and customizable graphic GUI, selectable dual mono stereo mid and side operations. A drive function that controls analogue saturation, selectable internal or external sidechain with filtering and equalization, a pre post function for the tone equalizer that can be placed before or after the compressor, a three position beef switch with thin, neutral and thick settings that change the compressor's response to an input signal, a loud relief function that relaxes the response to transients during heavy compression of loud program material. Allowing a natural dynamic response with extreme settings and amid side panel that can be used as a simple stereo widener, or fix any mid side level discrepancies. Amid or side mode, I have set up the compressor to compress a pretty boring and safe drumbeat. I've set the compressor to receive an external side che trigger, which in this example is another drumbeat. This trigger shapes the compressor's behavior and allows for a rhythmic response with exaggerated transients. I have selected the gritty model because my aim is to make the response dirty and thick. And I've supplemented this with a modest drive input value of six decibels, which affords gentle saturation. I've used an attack setting of 10. 2 milliseconds, which allows the initial beat transients to come through unaffected. A release value of 190 milliseconds allows the compressor to clamp down on the audio, compress and then release in between the transients. It's best to fine tune the release to taste and I've deliberately opted for a choppy type of effect. Threshold is at minus 44.5 decibels. This allows most of the audio to be compressed as opposed to just the peak transients. If the threshold is set too low, the audio could come across as too breathy and pumping. Adjust this by Yeah. To achieve the effect you're after. Ratio is set to 30 to 1 and loud relief is at 88 percent which tames some of the heavy transients generated by the low threshold and heavy ratio. Beef is set to thick and the knees are halfway compromised and set to 10 decibels. A hard knee is too extreme in this instance and a soft knee doesn't highlight the transients enough so a compromise was agreed on. Yes, all compressors have feelings. Anger them at your peril. Finally, mix the parallel to taste making sure not to clip the signal. I've supplied the fully wet file above so you can gauge the wet signal in isolation. The result is a transient, pronounced, and grimy response, which has added an interesting dimension to a state drumbeat. In example two, I'm using Fab Filters ProMB, which is a multi-band compressor, to raise and highlight the quiet parts of a s synth line. Upward compression is excellent at raising low level signals and is very effective in pronouncing the quiet parts of spoken vocals, drum beats, and so on. Here is the dry version. And now with the Pro MB. FabFilter's Pro MB is one of those plugins that doesn't just concentrate on multi band compression but offers expansion as well and offers it in a simple and detailed processor. When you create a band you can specify whether you want that particular band to behave as a downward or upward compressor or as a downward or upward expander. This allows the user to use this processor for finely detailed audio manipulation. As you would expect, ProMB comes with all the usual compressor expansion features. Threshold, Ratio, Tack and Release, Look Ahead, Knee, Linear, Minimum and Dynamic Phase Modes, Oversampling and Band Specific Side Chaining, Middle or Side Only Processing and a Parallel Mix Function. However, where the clever stuff takes over is in the way a simple control called deals with all four modes of compression and expansion. When selecting compression, a negative range setting results in downward compression, whereas a positive range setting results in upward compression. When opting for expansion, a negative range setting results in downward expansion, whereas a positive range setting results in upward expansion. But what makes this particular function so powerful is that it is variable and not fixed. This allows for gentle to extreme processing. For this particular example, I've used dynamic phase, which is a happy compromise between linear and minimum phase modes. In effect, it removes pre ringing artifacts and negates the latency that would normally be introduced in linear phase mode. I've created a band of 1.673 kilohertz and adjusted the bandwidth using the mouse wheel. A threshold of minus 20 decibels, a positive range of plus 15 decibels, attack at zero, and release at six percent. FabFilter like to use a percentage value as opposed to a time value for this particular processor, where zero percent is fastest and 100 percent is slowest. A ratio of 10 to 1 and a midway knee value of 24 decibels. The result is a well defined synth line that has all the low level transients boosted to match the louder transients. The synth sounds fuller and more defined. In example 3, I'm using Presonus Studio One built in downward expansor to effectively remove the low level sounds within a drum beat so as to leave just the kick and snare. Here is the dry version and now with the expander. Presonus have supplied users with an arsenal of very useful processing tools and the bundled expander is a simple yet powerful dynamic. To find the exact point at which a sound needs to be attenuated it's best to start with the expander already set to work and then to adjust the threshold to achieve the desired effect. I always start with a midway range setting, a hefty ratio and with fast attack and release times. And then adjust the threshold until I see and hear the expander working. For this particular example, I opted for a minus 16.7 decibel threshold. A 34 decibel range, setting a ratio of 20 to one with an attack of 10 milliseconds and a release of 300 milliseconds. An engaged look ahead, which is set at two milliseconds, look ahead is a useful function as it allows the expanded to start reacting up to two milliseconds before game change is actually detected. This is an excellent way to preserve transients while still avoiding ultra fast attack times that might cause distortion or aliasing. The result is a drum beat with no hi-hats evident, but with the kick and snare unaffected. This is a very cool way of removing low level sounds from a drum beat to leave only the primary sounds, which are usually the kick and snare. In example 4, I'm going to use the same drum beat, but this time run it through three different compressor topologies to demonstrate how each imparts its own character on the sound. Here is the dry version, and first with the PreSonus FET compressor, and then the PreSonus tube compressor, And finally with the IK Multimedia Dyna MU Compressor. I'll briefly run through the settings I use for each topology. The FET Compressor. I'm using the Presonus FET Compressor which as can be seen is an emulation of the classic URI 1176 FET Compressor. I've used a generous amount of input, minus 21dB. To drive the signal into the processor, attack is set to 0.52 milliseconds, which allows transients to pass through unaffected, and a release of 328 milliseconds, which I've made sure to time with the peaks and troughs of the signal. For ratio, I have selected the famous all mode, which affords a punchy and crunchy response. Finally, and this is really important when deciding on how the compressor behaves, fine tune the key filter. I have set this to 1.29 kHz which excludes frequencies below this value to reach the compressor's detector. The result is a tighter drum beat with a more emphasized kick and snappier snare. Now the tube compressor. I've opted for another PreSonus Dynamic and this particular compressor is an emulation of the famous Telectronics LA 2A tube driven opto compressor. This is one of my favourite glueing and colouring tools and it is deceptively easy to use. It comes with a simple limiter compressor toggle switch, an input gain which is critical in driving the signal into the processor and achieving lovely vowel joyfulness, and a peak reduction knob which sets the amount of compression. I've set the input gain to 45 and peak reduction to 52. I've set the key filter to 297 hertz. This pretty much takes the kick element out of the detector and allows the body of the beat to trigger the compressor. The result is a punchy yet warm beat with a rounded kick and snappier yet sustained snare. Next up Dyna MU Compressor. IK Multimedia's DynaMU is the weapon of choice here. Based on the famous manly variable MU stereo limiter compressor, the DynaMU can sound smooth and warm but can also be driven for saturation joyfulness, affording a rich punchy texture. DynaMU offers both left and right and mid side processing and can behave as a compressor or limiter depending on which mode you select. Comes with link and gang features. Link addresses the sidechain detection mode whereas gang links both channels for matched left right processing. Ratio is not displayed but this processor defaults to a soft compression ratio of 1. 5 to 1. If hard mode is selected the ratio uses a medium value of 4 to 1. The sidechain has a fixed 100 hertz high pass filter in place and cannot be altered. As with all tube driven processors, it comes down to balancing the input drive against the threshold. I've selected an input value of 2.9 dB, which is actually quite strong, as input starts at minus infinity and ends on plus 15 dB. The threshold range is 0 to 10 and I've selected a value of 4. Attack is set to 3.6 and release is set to 3, which is bordering on very fast. Output is at 3.8 decibels plus. The result is a vibrant and punchy yet smooth drum beat. The interesting observation here is that even if you try to match all the settings between the various topologies, you will invariably get different responses. And that highlights how various designs affect the sound in subtle and extreme ways. Understanding these designs is as important as understanding the workings of a compressor. In example 5, I'm going to create a couple of reverb effects using the sidechain of a compressor to duck the reverb on a clap sound. However, the technique I'm using involves nudging the sidechain trigger to reshape the compressor's response on the effect channel. I'll explain in detail how this works. Here is the dry version. And now with two versions of processing using Eventide's Ultra Reverb and FabFilter's Pro C2 Compressor. I'm using a single clap hit that has been duplicated and put on two separate channels in Presonus Studio One. I've disabled the main output of the copied clap as this will act as a sidechain trigger. In other words, a ghost trigger. And we route it to the sidechain input of FabFilter Pro C2 Compressor. This copy will also be nudged within Studio One using the provided Nudge tool. I settled on a sample nudge of 2, 646 samples, but please feel free to alter this to taste. The important aspect of this exercise is to make sure that the Nudge tool nudges in sample increments and not bars or beats. In Studio One, the Sample Nudge option is available via the Timebase menu option. I'm using the wonderful Eventide Ultra Reverb plugin, which is placed before the compressor on its own effects channel. The compressor is being triggered by the copied clap, i. e. the ghost trigger, and it is this sound that will be nudged. Select whatever you want for the reverb, but a long, dense reverb works really well with this exercise, and I'm using the basic chamber preset. I have shaped the compressor to swell the reverb effect by using the following settings. Threshold is at minus 60 dB for the late burst effect and minus 25 dB for the smoother response. The late burst effect uses the nudged clap trigger. Ratio is at 100 to 1 which allows for the whole reverb effect to be processed. Attack is at minimum which is 0. 005 milliseconds. Release is at 11 milliseconds which allows the compressor to swell when triggered. Knee is midway and the auto gain feature has been enabled. The external sidechain has no filtering as the whole clap response is being used to trigger the compressor. You can adjust any of the above to taste and I recommend that you do. You will be surprised at how the response of the reverb changes with micro adjustments to the compressor. In effect, no pun intended, the compressor is acting as a shaping tool for the reverb response. And the varying nudge values allow for different responses to be created. Experiment with nudge values, compressor and reverb settings as you will be quite surprised at how easy it is to create a variety of textures using this one simple technique. In example 6, I'm using FabFilter's ProMB to balance and highlight vocals. Here is the dry version. Such delight, as prison birds must find in freedom, Winging wildly across the white orchards and dark green fields. And now the processed version. Everyone suddenly burst out singing, And I was filled with such delight As prison birds must find in freedom Winging wildly across the white orchards And dark green fields. I have never been a fan of huge cuts or boosts using an equalizer on vocal lines. The response always sounds a little unnatural to my ears and trying to manage the crossover slopes between filter types can lead to all manner of problems. A drastic slope can be unnatural when filtering vocals, whereas a smooth slope can help to get the vocals to sit more naturally in the mix. Again, this comes down to taste and subjectivity, but there does come a time whereby the integrity of the signal will suffer, and this is usually a good time to rethink shaping the overall EQ response. I find there is a better way of achieving a more natural response and that entails working the frequencies around the selected band in unison with the band itself. If you dip before and or after the target frequency while gently boosting the target, then the boost sounds more pronounced and natural than simply boosting the specified target by a higher value. This is the technique I have employed in this example. To achieve this desired effect, I'm using FabFilter's ProMB, which allows me to use downward and upward compression and expansion independently of each other and simultaneously on the same file within a single processor. I've created two bands, one at 562 Hz and utilizing downward compression, and another at 4. 7 kHz and utilizing upward expansion. Both bands overlap and use 24 dB per octave slopes. The settings for each are as follows. Band 1 is using the compress mode with the threshold set to minus 31 dB. Range is minus 6 dB. A negative value denotes downward compression. Attack and release are both set at 20 percent. Knee is midway. And ratio is at default, 4 to 1. Band 2 is using expansion mode with the threshold set to minus 47 dB. Range is plus 4. 17 dB. A positive value denotes upward expansion. Attack and release are both set at 20%. Knee is midway. and ratio is at default 4 to 1. Every time band 2 upwardly expands, band 1 downwardly compresses, giving the impression that the expanded band is actually boosted. When both bands do not achieve the set threshold, they drop back down to flatline or zero. This gives a far smoother response with an undulating response that lends itself nicely to add clarity, balance, and motion to the vocal line. In the seventh and final example, I'm going to use a bunch of Cubasis supplied loops and make them a tad more interesting using sidechain compression, I have exaggerated the vocal reverb and synth hook so you can hear how the compression and expansion processes lend themselves to shaping texture and motion. Here is the dry version. And now the bells and whistles processed version. Okay, so a bunch of things are happening here and at my age, I have to make notes just to keep up. I'll start with the synth hook. I'm using FabFilter's Pro MB and this time I'm using it in middle and side mode. I've created two bands, one that incorporates downward compression and is set to mid, and the other that utilizes upward expansion on the sides. Both bands are being triggered by the drum beat via the sidechain. Settings are as follows. Band 1 is set to downward compress at 349 Hz. Threshold is minus 28 dB. Range is minus 9 dB. Attack and release are at 20 percent with the knee at 24 dB and ratio 4 to 1. This band is set to mid. Band 2 is set to upward expand at 1 kHz. Threshold is at minus 29. 6 dB. range is plus 5. 6 dB, attack is at 7. 6 percent and release is at 9. 8 percent with the knee at 24 dB. and ratio 4 to 1. This band is set to sides. As band 1 downward compresses, band 2 upward expands, giving that cool undulating motion effect. The fact that band 1 is set to mid keeps the low frequencies in check, but band 2 expands the sides, lifting the mid range frequencies, which results in the perception of there being more width at this particular range. For the vocal line I'm using both FabFilter Pro MB placed before Eventide's mighty SP 2016 reverb. I'm sidechain triggering the Pro MB with the same drum beat as I used with the synth hook. The Pro MB shapes the vocals and feeds directly into the SP 2016. I'm using Pro MB in middle and side mode with band 1 as before processing the mid and band 2 processing the sides. Settings are as follows. Band 1 is set to downward compress at 362.88 hertz, threshold is minus 33dB, range is minus 6dB, attack and release are at 20 percent with the near 24dB and ratio 4 to 1. This band is set to mid. Band 2 is set to upward expand at 3.13 kilohertz. Threshold is at minus 27dB, range is at plus 5. 92dB, attack and release are both set to 20 percent respectively, with the near 24dB and ratio 4 to 1. Band 2's output is raised by 2 dB. This band is set to sides. And now the SP 2016 reverb settings. Pre delay is at 19 milliseconds. Decay is 4.3 seconds. Position is set to 69%, which is closer to rear than front. Diffusion is at 85%, mix is set to 38%. These settings create a lovely bright reverb wash that hangs and dissipates between the sung words. This chain of configuration works really well in affording dynamic motion in both the vocal delivery and texture of reverb. To end this podcast I've highlighted the vocal line and the synth hook so you can gauge how much motion and richness is afforded by using compression, expansion and a reverb. That's it for now. Thanks for listening. This has been Eddie Bazil for Sound On Sound. All this love, All this love, Where is all my breathing at? In the air, In the air. Thank you for listening, and be sure to check out the show notes page for this episode, where you'll find further information, along with web links and details of all the other episodes. Oh, and just before you go, let me point you to the soundonsound.com/podcasts website page, where you can explore what's playing on our other channels.